HTML网页开发工具

• Outcome: Review
• Latest audit: Aug 10, 2026 · 11:22 AM UTC
• Version: 1.0.0

Overview

The skill is mainly an HTML review guide, but it requests command execution and write authority and includes broad API/file/command capabilities that are not clearly scoped to that purpose.

Review this skill carefully before installing. It appears to be markdown guidance rather than executable code, but it asks for exec and write tools and includes broad, generic automation capabilities. Use it only in an environment where command execution, file modification, and network/API access are restricted or require explicit approval.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Description-Behavior Mismatch

High

• Confidence: 96% confidence
• Finding: The skill is marketed as an HTML review helper, but later broadens itself to file processing, API integration, command execution, and information retrieval. This creates a capability mismatch that can mislead users and downstream agents into granting or invoking far more powerful operations than the stated use case requires, increasing the chance of unsafe execution or data exfiltration.

Context-Inappropriate Capability

High

• Confidence: 98% confidence
• Finding: Declaring command execution for a skill whose stated purpose is HTML mistake detection violates least privilege. If an agent or user supplies untrusted HTML, URLs, or derived parameters that influence shell commands, this expands a simple review workflow into a potential command-injection or arbitrary execution path.

Context-Inappropriate Capability

Medium

• Confidence: 88% confidence
• Finding: Documenting external API integration for a nominally local HTML review skill introduces unnecessary network exposure. User-supplied HTML or URLs could be sent to third-party services, enabling unintended data disclosure, SSRF-like fetch behavior, or privacy violations if the network behavior is not clearly bounded.

Intent-Code Divergence

Medium

• Confidence: 84% confidence
• Finding: The skill's security section says user-uploaded HTML should be avoided to prevent XSS, yet the skill's primary workflow explicitly asks users to provide HTML for analysis. This contradiction signals that the skill's threat model is incomplete and may cause implementers to process hostile HTML unsafely in previews, logs, browser renderers, or report generation.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The skill advertises write, API, and command-execution behavior without clear, prominent warning to users that it may modify files, contact external services, or run system commands. This weak transparency can lead to unsafe consent, overbroad agent autonomy, and accidental high-impact actions beyond expected HTML review.

VirusTotal

VirusTotal findings are pending for this skill version.

Static analysis

No suspicious patterns detected.